Intelligence Theories and Methods

377 Urban Collaborative Innovation Network and Its Influencing Factors of the AI Field in the Yangtze River Delta Region Hot! Wang Yuefen, Zhou Hongyu, Cen Yonghua DOI: 10.3772/j.issn.1000-0135.2024.04.001 This study explores the mechanism and influencing factors of urban collaborative innovation, and promotes the development of science & technology innovation between regions and multiple subjects. It collects patent data in artificial intelligence (AI) in the Yangtze River Delta region from 2016 to 2021, combining cities and patented technology knowledge to build a collaborative innovation network based on 27 cities. It analyzes the network's centrality, cohesive subgroup, and structural hole using social network analysis. Additionally, exponential random graph models (ERGM) is used to analyze the influencing factors of urban collaborative innovation by combining the historical statistical indicators, city level, subordinate province, and historical experience network. We find that the urban collaborative innovation network in AI in the Yangtze River Delta region has increased in scale and stability over time, and has become balanced. Regarding the influencing factors of the network, the nodes' main effect plays an obvious role in promoting development and education expenditure at the industrialization level. Moreover, a homogeneous tie that affects the subordinate province and administrative level has different effects on network development. Additionally, the network has a relatively obvious path-dependence trend, and the actual network of the preceding year has an important impact on the network formation in the following year. 2024 Vol. 43 (4): 377-390 [Abstract] ( 134 ) HTML (196 KB)  PDF (8476 KB)  ( 138 )

391 The Influence of Network Characteristics of Cross-Border Teams on Disruptive Innovation Performance Hot! Lin Chunpei, Zhu Xiaoyan, Yu Chuanpeng, Liao Yangyue, Li Hailin DOI: 10.3772/j.issn.1000-0135.2024.04.002 Cross-border teams play an important role in the disruptive innovation activities of innovation entities such as enterprises, and the application of machine learning methods to identify the configuration path between their network characteristics and disruptive innovation performance is an important problem to be solved. Based on 139,999 patent data in the UAV field of the Incopat patent search platform, this study uses the community discovery algorithm to identify 185 cross-border teams from the cooperation relationship data of patent inventors, selects the network characteristic variables of cross-border teams according to social network theory, and uses the k-means clustering algorithm to classify cross-border teams. Furthermore, we used the decision tree CART algorithm to explore the influence of different types of cross-border team network characteristics on disruptive innovation performance. The results show that (1) there are three types of cross-border teams: binary cooperation, quasi-perfect cooperation, and complex cooperation, and different cross-border team types have different effects on disruptive innovation performance; the quasi-perfect cooperation team has the highest proportion of highly disruptive innovation performance, while the dualistic cooperation team has the lowest proportion of highly disruptive innovation performance; (2) cooperation intensity is universal, which is the core factor that affects the disruptive innovation performance of different cross-border teams at different levels; and (3) cooperation intensity positively affects the disruptive innovation performance of binary cooperative teams. The disruptive innovation performance of quasi-perfect cooperative teams is jointly affected by the aggregation coefficient, cooperation intensity, and team size. For complex cooperative teams with high cooperation intensity, maintaining a low network density is conducive to improving disruptive innovation performance. 2024 Vol. 43 (4): 391-404 [Abstract] ( 79 ) HTML (163 KB)  PDF (2956 KB)  ( 188 )

405 Recapturing the Flow of Knowledge: Tracing Structured Knowledge Back to Historical Records Hot! Zhang Qi, Kong Jia, Hu Haotian, Wang Dongbo, Wang Hao, Deng Sanhong DOI: 10.3772/j.issn.1000-0135.2024.04.003 Tracing structured historical knowledge back to historical records can enhance the verifiability and reliability of knowledge. In response to the challenges of inadequate knowledge tracing mechanisms in existing knowledge bases of ancient books and the absence of trigger words in several Archaic Chinese texts, this study introduces a method to trace structured historical knowledge back to historical records. First, a structured historical knowledge tracing framework is proposed by leveraging techniques such as co-reference resolution and textual entailment. Subsequently, a dataset is proposed to compare the effectiveness of different pre-trained language models, including BERT, SikuBERT, GPT-3, and GPT-4. This dataset combined with different input strategies on the knowledge tracing effect, is used to structure the historical knowledge tracing model, SHK-Tracer, which was employed to trace the historical subject matter knowledge base (Shiji Mutil-dimensional Knowledge Base, SMKB) to different ancient historical books, with 80.19% precision. We found that the knowledge overlap between Shiji and each historical fragment in historical books, such as Zuozhuan and Guoyu, did not correlate proportionally with the inherent information content of the historical fragment. The results of the study serve the dual purpose of first, aiding scholars and readers in verifying the authenticity of knowledge, by providing cross-references between different historical sources and identifying the original source; and second, facilitating digital humanities research, including historical knowmetrics, relation extraction, and linguistic style calculations of ancient historical records. 2024 Vol. 43 (4): 405-415 [Abstract] ( 83 ) HTML (120 KB)  PDF (2098 KB)  ( 85 )

416 A Text Semantic Matching Model Based on Knowledge Enhancement Hot! Zhang Zhengang, Yu Chuanming DOI: 10.3772/j.issn.1000-0135.2024.04.004 Text semantic matching models have been widely used in information retrieval, text mining, and other fields. Current methods mainly predict the semantic relationship of text pairs from the perspective of the text itself, ignoring external knowledge. We propose a new text semantic matching model based on knowledge enhancement to address this issue. The model utilizes knowledge graph entities as external knowledge, effectively models the text’s external knowledge information, and adaptively filters the noise in the external knowledge. Our model achieves the best results for most indicators in Natural Language Inference and Paraphrase Identification. This research will aid in the application of knowledge graphs in text semantic matching tasks and provide a reference for applying knowledge graphs to the information field. 2024 Vol. 43 (4): 416-429 [Abstract] ( 75 ) HTML (308 KB)  PDF (1863 KB)  ( 202 )

430 Decision-Oriented Map Knowledge Service Hot! Qi Xiaoying, Li Xinwei, Yang Haiping, Xu Panqing, Lu Lan DOI: 10.3772/j.issn.1000-0135.2024.04.005 To integrate and transform the geopolitical, historical, and cultural information in maps into intelligence for decision-making, based on the interactive combination of “long-term” and “process geography,” this paper considers the synchronicity of the spatial field and the diachronic nature of the time order, to construct a map knowledge service system for a practical study of the South China Sea. First, through the five intelligence analysis capabilities comprising discovery, evaluation, interpretation, prediction, and delivery, a decision oriented map knowledge service system is constructed, and the key issues and core links of the knowledge discovery mechanism, consisting of evaluation, association, reasoning, and delivery mechanisms in the system are explained. Second, through the knowledge service system map, the South China Sea map database and map classification system, metadata framework and map ontology model, knowledge graph, question answering system, evidence chain, and other knowledge service practices were established considering the four aspects of the South China Sea map: resource construction, semantic organization, knowledge association, and knowledge delivery. The paper provides a powerful theoretical basis and intelligent tool promoting the intelligence value of the South China Sea map in the decision of “evidential war.” 2024 Vol. 43 (4): 430-445 [Abstract] ( 87 ) HTML (131 KB)  PDF (4468 KB)  ( 99 )

446 Cross-Social-Media Public Opinion Risk Perception: Construction and Implementation of a Theoretical Framework Hot! Wang Dan, Liu Fukang, Lu Wei DOI: 10.3772/j.issn.1000-0135.2024.04.006 The shifting and moving among social media platforms allow Internet users to obtain complex and diversified information, accelerate the efficiency of public opinion communication, and accelerate the generation and evolution of public opinion risks. Timely capture and prediction of online public opinion risks are crucial for maintaining network security. Currently, the perception objects of public opinion risk mainly focus on a single social media platform, and there is room to expand the construction and implementation of a theoretical framework for cross-social-media public opinion risk perception. Therefore, by analyzing the communication characteristics of cross-social-media public opinion risk, this study systematically develops a cross-social-media public opinion risk perception framework that includes three stages: identification of abnormal public opinion on a single media platform, research and evaluation of cross-social-media public opinion risk, and prediction of cross-social-media public opinion risk. Through the construction of a multistage index system and data correlation mining, the identification, research, and prediction of cross-social-media public opinion risks are realized. The findings can broaden the research perspective of public opinion risk perception, enrich the theoretical system of public opinion risk, and strengthen the synergy of social media to enable public opinion risk governance, which will be conducive for improving the level of public opinion risk prevention and control. 2024 Vol. 43 (4): 446-456 [Abstract] ( 171 ) HTML (167 KB)  PDF (1711 KB)  ( 143 )

457 Factors that Influence the Dissemination Effects of Short Videos of Refuting Rumors Based on Heuristic-Systematic Model Hot! Fu Shaoxiong, Su Yiqi, Sun Jianjun DOI: 10.3772/j.issn.1000-0135.2024.04.007 The key to refuting rumors on short video platforms is to enhance the dissemination of short videos refuting rumors. To improve the dissemination effect of short videos refuting rumors and deepen the ecological governance of network information, we analyzed the influence of content design and posting techniques for short videos refuting rumors on their dissemination. This study manually coded 965 official disinformation videos on TikTok. Based on the heuristic-systematic model, heuristic (posting techniques) and systematic (central and peripheral content) cues were contextualized within a short video platform, and the number of likes, comments, favorites, and shares were used as indicators of dissemination effectiveness to analyze the effect of heuristic-systematic cues on dissemination. The results of the regression analysis indicate that the number of likes, topic types in heuristic cues, information completeness, information uniqueness, title symbols and modal diversity in systematic cues had significant effects; for the number of collections, duration, posting time period, topic types and background music volatility in heuristic cues, information completeness, information uniqueness, title styles, and title symbols in systematic cues had significant effects; for the number of comments, posting time period and background music volatility in heuristic cues, information uniqueness and title symbols in systematic cues had significant effects; for the number of shares, posting time period and the linkage in heuristic cues, information uniqueness and the way of information presentation in systematic cues had significant effects. This study extends the research perspective of short videos refuting rumors, expands the research context of the heuristic-systematic model, and clarifies the factors that influence the dissemination effects of short videos refuting rumors. 2024 Vol. 43 (4): 457-469 [Abstract] ( 95 ) HTML (213 KB)  PDF (1179 KB)  ( 151 )

Intelligence Technology and Application

470 Research on Literature Classification Methods Based on Multiple Feature Correlation and a Graph Attention Network Model: A Case Study of Chinese Medical Literature Hot! Chen Shuaipu, Qian Yuxing, Qian Zhiqiang, Liu Zhenghao, Zhang Zhijian DOI: 10.3772/j.issn.1000-0135.2024.04.008 The increasing complexity and detailed themes of scientific literature pose a significant challenge for efficient classification. A potential solution is the development of automatic literature classification technology, enabling the intelligent management of information resources and efficient scientific research retrieval. In response, this research presents a Hierarchical Text Classification Networks based on Multiple feature Correlation and Graph Attention Network (HTCN-MCGAT) to overcome the limitations of traditional methods. The HTCN-MCGAT model comprises three integral components: (1) The text representation and enhancement module redesigns the fine-tuning stage of the Bidirectional Encoder Representation from Transformers pre-training model to enhance the representation of the current literature at two levels: the internal character correlations of literature abstracts, titles, and keywords, and external document correlation; (2) The label association modeling module employs the Graph Attention Network to model the hierarchy and relationships between label semantics; and (3) The hierarchical interaction classification module incorporates a hierarchical fusion attention mechanism and a hierarchical classification framework that consists of global and local information based on multi-task learning for integrating high-level features classification. The proposed model is applied to the Chinese medical literature domain and tested with a series of experiments. The results demonstrate the HTCN-MCGAT model’s superior performance compared to traditional literature classification methods, improving the F1-score by 4.34%-13.21%. This research offers an optimized approach to literature classification from text-semantic enrichment and hierarchical-relationship-modeling perspectives. The findings hold potential for applications not only in literature classification tasks but also in hierarchical classification fields. 2024 Vol. 43 (4): 470-490 [Abstract] ( 80 ) HTML (309 KB)  PDF (2637 KB)  ( 128 )

491 Bias Identification from an Evidence-Based Perspective: A Big Data Meta-Analysis Procedure Based on Egger's Extension Model Hot! Zhou Wenjie, Lin Weijie, Wei Zhipeng, Yang Kehu DOI: 10.3772/j.issn.1000-0135.2024.04.009 Research synthesis serves as a bridge between academic research findings and the development of practice guidelines. As a tool for evidence integration and translation, meta-analysis is at the core of evidence-based practice. However, the reliability of meta-analysis results is often compromised by biases. Addressing the common issues of selection bias and outcome reporting bias in the process of evidence synthesis, this study aims to extend the model developed by Egger and others through meta-regression. It employs a mathematical decomposition method to effectively identify selection bias and outcome reporting bias, thus developing a new approach for bias identification. Building upon the establishment of an accurate bias identification extension model, this study further validates the rationality and scientific validity of the extended model using a set of empirical research data. The developed extension model significantly enhances the efficiency of Egger’s test, contributes to improving the quality of meta-analysis, and aids in the construction and refinement of a scientific evidence-based social science theoretical framework. 2024 Vol. 43 (4): 491-502 [Abstract] ( 81 ) HTML (204 KB)  PDF (780 KB)  ( 62 )